Health care expenditures account for 17.3% of gross domestic product in the United States and are disproportionately allocated to care at the end-of-life (EOL). Much of this results from intensive use of services (hospitalizations, mechanical ventilation) in the last months of life. Yet, data suggest that terminally-ill patients who receive aggressive EOL care have worse quality of life than other patients. Research is needed to determine the factors that contribute to the intensity of EOL care. A large body of evidence demonstrates substantial regional variations in intensity of care and health care spending at the EOL across the U.S. Studies have also demonstrated notable differences in EOL care by race/ethnicity and health systems. Available data have not had sufficient detail, variation, or siz to assess the extent to which area variations are explained by patient, physician, and health system factors, nor to assess if racial/ethnic disparities in EOL care can be explained by differences in patients' beliefs and other. characteristics, physicians' practice styles, the hospitals where care is received, or area practice patterns. We will use data from the Cancer Care Outcomes and Research Surveillance (CanCORS) Consortium, a multi-regional prospective study examining care delivered to population and health-system based cohorts of more than 10,000 patients diagnosed with lung or colorectal cancer during 2003-2005. We will use CanCORS patient survey data, medical record data and physician survey data linked with administrative data from Medicare, private health plans, Medicaid, and the VA to examine the intensity of EOL care among over 4,000 patients with advanced lung or colorectal cancer followed through 2012. We will use hierarchical models to assess the patient, physician, hospital, and area factors influencing intensity of EOL care. Specifically, we will: 1. Validate th retrospective Dartmouth measures of EOL spending with measures of care intensity and expenditures for prospectively identified patients with advanced cancer. 2. Understand the factors contributing to area-level variations in intensity of EOL care, including patient and tumo characteristics (e.g., demographics, comorbid illness, site of metastases), patient beliefs (e.g., preferences for life-prolonging care, beliefs about the effectiveness of chemotherapy for advanced cancer), physician practice style and beliefs (e.g., greater use of chemotherapy for advanced cancer patients, self- reported timing and comfort with EOL discussions, personal preference for hospice if terminally-ill), hospital characteristics and practice patterns, and are service availability. 3. Within areas, assess to what extent racial/ethnic differences in intensityof EOL care are explained by patient, physician, and hospital differences. 4. Understand differences in intensity of EOL care across health system (fee-for-service Medicare, Medicare managed care, VA), and assess patient and physician characteristics and beliefs across systems. PUBLIC HEALTH RELEVANCE: This study will expand our understanding of the factors contributing to the high expenditures and intensity of end-of-life (EOL) care for individuals with advanced cancer and will help to identify the appropriate targets for strategies to address disparities in EOL care. If differences in EOL care intensity are primarily driven by physicians, local area practice patterns, and availability of services (e.g., intensive care beds), the finding will identify opportunities for interventions to assure that patients obtain care at the EOL that matches their preferences, and it will identify opportunities to decrease intensity of care in area where such care is not driven by patients' preferences. If high-intensity EOL care is driven by patients' preferences, then it will be important to assess if these are informed preferences and, i not, whether they can become informed preferences (e.g., preferences that may be modifiable through EOL discussions).